Impact of alternative initiation, splicing, and termination on the diversity of the mRNA transcripts encoded by the mouse transcriptome

Abstract

We analyzed the FANTOM2 clone set of 60,770 RIKEN full-length mouse cDNA sequences and 44,122 public mRNA sequences. We developed a new computational procedure to identify and classify the forms of splice variation evident in this data set and organized the results into a publicly accessible database that can be used for future expression array construction, structural genomics, and analyses of the mechanism and regulation of alternative splicing. Statistical analysis shows that at least 41% and possibly as much as 60% of multiexon genes in mouse have multiple splice forms. Of the transcription units with multiple splice forms, 49% contain transcripts in which the apparent use of an alternative transcription start (stop) is accompanied by alternative splicing of the initial (terminal) exon. This implies that alternative transcription may frequently induce alternative splicing. The fact that 73% of all exons with splice variation fall within the annotated coding region indicates that most splice variation is likely to affect the protein form. Finally, we compared the set of constitutive (present in all transcripts) exons with the set of cryptic (present only in some transcripts) exons and found statistically significant differences in their length distributions, the nucleotide distributions around their splice junctions, and the frequencies of occurrence of several short sequence motifs.

Figure 1

Distribution of distances between alternative splice sites: (A) 5′-splice sites; (B) 3′-splice sites. Note that we only show the distribution up to a distance of 100 nt.

Figure 2

Estimated frequency of lengths of constitutive (black) and cryptic (red) exons. For each value of the length L we determined the frequency in our data set of exons with length between 0.8 L and 1.2 L.

Figure 3

Nucleotide distribution in the (A,C) 5′- and (B,D) 3′-splice signals flanking constitutive (upper panels) and cryptic (lower panels) exons. The relative sizes of the letters indicate the relative frequencies of the nucleotides at that distance from the splice junctions. The absolute sizes of the letters correspond to the information score of the nucleotide distribution at that position.

Figure 4

Position-specific probability that the nucleotides at (A) 5′- and (B) 3′-splice signals of cryptic exons were generated from different versus the same underlying distribution as those of the constitutive exons.

Figure 5

Splice variants of polypyrimidine-tract-binding protein.

Figure 6

Annotation of variant exons. The splice sites responsible for the annotation of the associated exons as variant are indicated by vertical bars. Exons with invariant splice sites are shown in green, exons with alternative 5′ sites are shown in yellow, and those with alternative 3′ sites in blue. Cryptic exons are indicated by a black box surrounding the exon. Introns are shown in red.